Google open-sources TorchTPU so PyTorch apps can run on TPUs
Original title: Google has joined Amazon in attacking Nvidia’s stronghold with AI app developers
Amazon released TorchNeuron a day earlier to port PyTorch apps to Trainium without rewriting them.
Google said Tuesday that TorchTPU, its PyTorch version for Tensor Processing Units, is available as open source. A configuration change can move existing PyTorch apps to TPUs and Google Cloud without rewriting. Amazon on Monday released TorchNeuron for Trainium and AWS and said it is submitting code to decouple PyTorch features from Nvidia hardware.
Unconfirmed: The 85% PyTorch usage figure is presented only as “by some accounts.”
- The article says PyTorch is used by 85% of AI apps, by some accounts.
- Google says it worked with Meta, Hugging Face, Inferact, and RadixArk.
- Google launched a TPU agent suite and updated LiteRT for PyTorch apps.
- Meta transferred PyTorch to the Linux Foundation in 2022.
- Organizations
- Linux FoundationHugging FaceInferactRadixArk
From the original article
2 min read at techcrunch.com
Google has joined Amazon in releasing a developer tool that makes it easy for PyTorch users to move their apps off Nvidia GPUs.
By Julie Bort · techcrunch.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.